In [134]:
import numpy as np
import pandas as pd
import os
import requests
import pprint
import matplotlib.pyplot as plt
import requests
import operator
from datetime import datetime
from dateutil import tz
import operator
from collections import OrderedDict
from math import isnan
import json
import re
from IPython.display import Image
In [135]:
def wordcloud_plot(text, colour):
wordcloud = WordCloud(
width = 3000,
height = 2000,
background_color = colour,
stopwords = STOPWORDS).generate(str(text))
fig = plt.figure(
figsize = (40, 30),
facecolor = 'k',
edgecolor = 'k')
plt.imshow(wordcloud, interpolation = 'bilinear')
plt.axis('off')
plt.tight_layout(pad=0)
return plt.show()
In [136]:
def clean_tweet(tweet):
return ' '.join(re.sub(r"(@[A-Za-z0-9]+)|([^0-9A-Za-z \t])|(\w+:\/\/\S+)", " ", tweet).split())
In [137]:
path = os.getcwd()
path += '/incindia.json'
In [138]:
tweets = ""
with open(path, 'r') as f:
for line in f.readlines():
tweets += line
In [139]:
data = json.loads(tweets)
In [140]:
mostLiked, likes, likeurl = 0, 0, ""
mostComment, comment, commenturl = 0, 0, ""
for post in data['GraphImages']:
if likes < post['edge_media_preview_like']['count']:
mostLiked = post['id']
likeurl = post['shortcode']
likes = post['edge_media_preview_like']['count']
if comment < post['edge_media_to_comment']['count']:
mostComment = post['id']
commenturl = post['shortcode']
comment = post['edge_media_to_comment']['count']
Most Liked Post with screenshot¶
In [141]:
print("Post has the most likes' ID is: ",mostLiked)
print("Total Likes: ",likes)
In [142]:
likeurl
Out[142]:
In [143]:
Image("screenshot/insta-inc1.png")
Out[143]:
Most commented post with screenshot¶
In [144]:
print("Post has the most comments' ID is: ",mostComment)
print("Total Comments: ",comment)
In [145]:
commenturl
Out[145]:
In [146]:
Image("screenshot/insta-inc2.png")
Out[146]:
In [147]:
df = []
for post in data['GraphImages']:
df.append([int(post['edge_media_preview_like']['count']), int(post['edge_media_to_comment']['count']), int(post['id'])])
In [148]:
df = pd.DataFrame(df, columns=['likes', 'comments', 'id'])
In [149]:
newdf = df.sort_values(['likes', 'comments'], ascending=[False, False])
Top 5 most popular posts¶
In [150]:
newdf.head()
Out[150]:
In [151]:
hashtags = {}
for post in data['GraphImages']:
if 'tags' in post and len(post['tags']) > 0:
for tag in post['tags']:
if not tag in hashtags:
hashtags[tag] = 1
else:
hashtags[tag] += 1
In [152]:
sorted_x = sorted(hashtags.items(), key=operator.itemgetter(1))
Top 5 Hashtags¶
In [153]:
sorted_x[len(sorted_x)-5:len(sorted_x)]
Out[153]:
PART 2¶
In [176]:
import json
from wordcloud import WordCloud, STOPWORDS
In [177]:
path = os.getcwd()
path += '/bjp4india.json'
In [178]:
tweets = ""
with open(path, 'r') as f:
for line in f.readlines():
tweets += line
In [179]:
data = json.loads(tweets)
In [158]:
count = 0
for post in data['GraphImages']:
if post['is_video']:
count += 1
In [159]:
def showPieChart(photoTweet, rest):
slices_hours = [photoTweet, rest]
activities = ['Photo in Tweet', 'Other']
explode = (0.012, 0.012)
colors = ['orange', 'coral']
fig1, ax1 = plt.subplots()
ax1.pie(slices_hours, explode=explode,colors=colors, labels=activities, autopct='%1.1f%%',
shadow=True, startangle=120)
ax1.axis('equal')
plt.show()
Pie chart representing the share of images in the tweets as compared to videos¶
In [160]:
showPieChart(count, len(data['GraphImages'])-count)
In [161]:
df = []
for post in data['GraphImages']:
text = post['edge_media_to_caption']['edges'][0]['node']['text']
df.append([int(post['edge_media_preview_like']['count']), int(post['edge_media_to_comment']['count']), int(post['id']), text])
In [162]:
df = pd.DataFrame(df, columns=['likes', 'comments', 'id', 'post'])
In [163]:
newdf = df.sort_values(['likes', 'comments'], ascending=[False, False])
In [164]:
newdf.head()
Out[164]:
In [165]:
tweets = newdf['post'][0 : 200]
In [166]:
stopwords = set(STOPWORDS)
stopwords.update(['Nan','Unknown','https','co','@', 'shri'])
In [167]:
cleanTweets = []
for i in range(len(tweets)):
cleanTweets.append(clean_tweet(str(tweets.iloc[[i]].iloc[0])))
Wordcloud of top 200 most popular posts'¶
In [168]:
wordcloud_plot(cleanTweets, 'white')
Reason¶
The most occuring words are:
- PM - Current Prime Minister of India is from this party.
- BJP - The name of the party, it is used for emphasis and glorification of the party.
- India - The country of the party, signifies Nationalist feelings of the party.
- Election - It's election time and Post are related to election.
In [180]:
hashtags = {}
for post in data['GraphImages']:
if 'tags' in post and len(post['tags']) > 0:
for tag in post['tags']:
if not tag in hashtags:
hashtags[tag] = 1
else:
hashtags[tag] += 1
In [181]:
hashtagCount = {}
for data in hashtags:
if not type(data) is str:
continue
if data not in hashtagCount:
hashtagCount[data] = 1
else:
hashtagCount[data] += 1
In [182]:
sorted_x = sorted(hashtags.items(), key=operator.itemgetter(1))
Top 5 Hashtags¶
In [183]:
print("Top 5 Hashtags : ")
sorted_x[len(sorted_x)-5:len(sorted_x)]
Out[183]:
PART 3¶
In [184]:
path = os.getcwd()
path += '/explore_2019-04-14-05-00-26.json'
In [185]:
tweets = ""
with open(path, 'r') as f:
for line in f.readlines():
tweets += line
In [186]:
data = json.loads(tweets)[0]
In [187]:
df = []
for post in data['posts']:
df.append([post['caption'], int(post['likes']['count']), int(post['views']), int(post['comments']['count']), post['url']])
In [188]:
df = pd.DataFrame(df, columns=['caption', 'likes', 'views', 'comments', 'url'])
Data Formate¶
In [189]:
df.head()
Out[189]:
In [190]:
newdf = df.sort_values(['views', 'likes', 'comments'], ascending=[False, False, False])
Top 5 Most Popular Posts with Screenshots¶
In [191]:
newdf.head()
Out[191]:
In [192]:
urls = newdf['url'][0:5]
In [194]:
Image("screenshot/insta-explore1.png")
Out[194]:
In [195]:
Image("screenshot/insta-explore2.png")
Out[195]:
In [196]:
Image("screenshot/insta-explore3.png")
Out[196]:
In [197]:
Image("screenshot/insta-explore4.png")
Out[197]:
In [198]:
Image("screenshot/insta-explore5.png")
Out[198]:
In [199]:
count = 0
for post in data['posts']:
if len(post['imgs']) > 1:
count += 1
In [200]:
count
Out[200]:
Pie-Chart¶
In [201]:
showPieChart(count, len(data['posts'])-count)
In [ ]: